Import models

This section describes various ways to get model checkpoints and code into Studios. In principle, do exactly the same that you would do on your laptop today.

Import within Studios

Use any of these methods to import models from the Studio web app.

Upload to Drive

Click on the Drive icon > Add data, select the checkpoint file and press upload.

The model will now be accessible from any Studio in the teamspace.

Upload to VSCode or Jupyter

Drag and drop the checkpoint file into the VSCode or Jupyter interface which will automatically upload the files.

The model will now be available to access from any Studio in the teamspace. For more on Lightning SDK, read our in-depth guide.

Import from Hugging Face hub

This example shows how to download a model from the Hugging Face hub via the Studio terminal.

Import from Github

Find the repository that has the model checkpoints and clone the repo.

Import using Studio terminal

Use the Studio terminal to download any file (like model checkpoints) using the curl command.

Import from outside Studios

Use these methods to upload models from outside Lightning AI programatically.

Upload via CLI

Use the Lightning SDK to upload the model checkpoint into the teamspace.